On The Scope Forward Show, Matt Schwartz—Founder and CEO of Virgo—unveiled EndoML, a clinician-facing platform built on Virgo’s massive foundation model, EndoDINO. This isn’t just another AI update—it’s a turning point for gastroenterology.
With more than 2 million full-length endoscopy videos and a platform that lets any clinician train models using their own technique, Virgo’s trajectory signals a fundamental shift—from industry-led tools to clinician-owned AI.
This isn’t an update. It’s a wake-up call.
Matt lays out a bold vision: Imagine any endoscopist building a CADx system tuned to their own hands, their own diagnostic eye. Picture AI not as a separate assistant, but as an extension of the clinician, scaling personal expertise across teams, institutions, even continents. And what if the real bottleneck in AI for GI wasn’t regulatory, but data—and that hurdle is already behind us?
EndoML isn’t just a tool—it’s a scaffolding layer for AI-native medicine. If you’re in GI and not exploring this shift, you may already be behind the exponential curve.
Watch the episode.
Top Insights from the Interview:
1. The World’s Largest Endoscopy Video Database
Virgo has captured over 2 million full-length HD endoscopy videos, with more than 1 million new procedures added annually. This massive dataset fuels the power of its AI.
2. EndoML: AI for Every Endoscopist
EndoML lets clinicians upload, label, and train AI models using their own data, cutting down the training volume needed to build high-performance models.
3. EndoDINO Enables Generalizability
Trained on 3.5 billion frames of real-world data, EndoDINO powers models that can generalize across diverse clinical environments.
4. AI as an Extension of the Endoscopist
Clinicians can now train AI to reflect their personal technique—transforming AI into a mirror of their expertise and enabling scalable, high-fidelity clinical replication.
5. From AI Models to Autonomous Robots
As EndoML captures clinical intent and expertise, robotic endoscopy becomes viable—AI not just assisting, but eventually performing under oversight.
6. The Exponential Fallacy: Why Most Will Miss the Curve
Matt highlights a recurring trap in tech: underestimating exponential growth. In GI, tools like EndoML are advancing faster than most realize.
7. Humanoid Robots in GI? It’s Coming
With advances in robotics like Tesla Bot and Figure, GI procedures may one day be performed by humanoid robots, powered by clinician-trained AI.
8. The WTF Curve of Innovation
Your first reaction might be disbelief. But just past that “WTF” moment lies opportunity for early adopters willing to lean in and shape the future.
9. AI in GI Is Moving Faster Than You Think
From foundational models to clinical tools, AI’s pace in healthcare is accelerating. The gap isn’t technical—it’s mindset. Those who delay may be left behind.